Planning an IVDR clinical performance study

Confirm the regulatory route

IVDR distinguishes different types of performance studies and sets conditions for application, notification, conduct, substantial modification, corrective measures and reporting. The exact route also interacts with Member State requirements and the study’s use of specimens, additional invasive procedures and clinical management decisions.

Before budgets and dates are fixed, write a route memo covering the study category, countries, competent-authority and ethics steps, sponsor responsibilities, insurance or indemnity needs, safety reporting and expected submission documents. Uncertainty here is a schedule dependency, not an administrative detail.

Test site and specimen feasibility

A site may see many relevant patients and still be unable to deliver the study. Feasibility should test the exact population, specimen type and volume, pre-analytical timing, reference method, data availability, consent route, competing studies and staffing needed for the proposed workflow.

For archived or residual specimens, verify provenance, storage history, freeze–thaw limits, data linkage and whether the available set reflects the intended-use population. For prospective studies, confirm the actual recruitment funnel rather than relying on annual disease counts.

Develop the data model alongside the analysis plan

The protocol, statistical analysis plan, eCRF and laboratory worksheets should be developed as one system. Every primary endpoint needs the variables, timing, units, coding and query rules required to derive it. Invalid, indeterminate, missing and repeated results need predefined treatment.

A traceable dataset is not created by the EDC alone. It depends on controlled source definition, role-based access, edit checks, query handling, monitoring, audit trails, version control, database lock and documented exports that match the analysis plan.

• Map each claim and endpoint to its source fields and derivation.
• Predefine analysis populations and protocol-deviation rules.
• Control units, reference ranges, comparator results and specimen identifiers.
• Test the end-to-end data flow before the first evaluable subject or specimen.

Control protocol and device changes

Changes to population, endpoints, specimen handling, device version, comparator, sites or analysis can affect participant protection, data reliability or the interpretation of results. Each change needs a documented impact assessment. A substantial modification may require notification or authorisation before implementation, except where immediate action is needed to protect subjects.

Safety reporting also needs a defined process. The sponsor should establish who identifies, assesses, documents and reports adverse events or device deficiencies, using the current Commission and MDCG guidance applicable to IVD performance studies.

Plan the study report before close-out

The Clinical Performance Study Report should be the planned endpoint of the evidence chain, not a retrospective writing exercise. Tables, listings, deviations, exclusions, device accountability, safety information and limitations should be traceable to approved plans and controlled data.

Before close-out, confirm who resolves outstanding queries, approves the analysis and reviews the final report.

Work backwards from the planned analysis

Take one primary endpoint and identify the records needed to calculate it. A diagnostic performance estimate may require the index test result, an appropriate reference classification, specimen eligibility and predefined treatment of indeterminate or missing results. If the collection form does not capture those inputs, the analysis cannot be repaired merely by choosing a different statistical method at the end.

The sample-size discussion should state the assumptions and the precision or hypothesis being addressed. Total specimen count alone can hide an insufficient number in a clinically important subgroup or outcome category. Agree the analysis populations and exception handling before data collection, and document later changes with their rationale and impact.

Ask sites for evidence of feasibility, not just enthusiasm

A site’s annual laboratory volume is a starting point. Ask how many specimens are likely to meet the actual eligibility, specimen handling and reference-data requirements in the planned period. Understand staff availability, the testing workflow, storage, competing work and the steps needed before activation. For a multi-site study, compare these factors using the same definitions.

Include operational exceptions in the feasibility discussion. Clarify who handles invalid runs, missing information, device accountability and queries. If testing and recruitment occur in different locations, the handover needs to preserve the information required for eligibility and analysis. An apparently large network is not a substitute for confirming that the selected sites can perform this protocol.

Define closeout while the study is being designed

Agree what is needed for a controlled database lock and final report: reconciled queries, documented deviations, traceable exclusions, safety information and approved analysis outputs. Decide who reviews the tables and how the results will feed into the performance evaluation report. These arrangements influence data collection from the start.

If the study produces a result outside expectations, report it with the appropriate analysis and limitations. Do not narrow the population retrospectively simply to make the headline result more favourable. A useful final report explains what was planned, what happened and what the evidence supports for the device’s intended purpose.